This project addresses statistical problems generated from collaboration with scientists in other program areas and general statistical problems of current interest. This project is a continuing activity of the Section on Mathematical Statistics. Papers have been published in FY'85 on the following statistical subjects: tests of monotone trend for k Poisson means, a diffusion approximation of absorption probabilities for binomial random walks in the presence of linearly varying boundaries, modeling covariate dependence of binary sequences, pooled adjacent violators, and general tests of trend for count data. Other work in progress includes the design and analysis of Phase II clinical trials, nonparametric tests with quadratic alternatives, statistics for the evaluation of space-time clustering of disease, modeling of residual treatment effect for k-period two-treatment crossover designs, statistical inference under inequality constraints, adjustments for covariates in the analysis of categorical data, the influence of missing data on statistical models determined by variable selection procdures, the effects of misclassification of exposure variables on case-control studies, the use of area surveys in epidemiological research, and sampling strategies for rare diseases.